Mat - thanks for your review and input. I specified the two bits only for backward compatibility with existing implementations.
I imagine we could design a specification that retains one bit and deprecates the other, with rules about the appearance of the depre backward compatibility. At least, this spec would need a note explaining why there are two bits in the spec and recommending that new implementations use, for example, just the M bit. - Ralph On Fri, 2005-05-27 at 13:41 +0100, [EMAIL PROTECTED] wrote: > Ralph Droms <[EMAIL PROTECTED]> wrote: > > > Seems to me I'm hearing two requirements out of this thread: > > > > 1) Ability to indicate to a host "DHCP is not available on this link", > > with the expectation that the host won't send any DHCP messages > > > > 2) Ability for a host to get all desired and available DHCP > > configuration with a single DHCP message exchange > > - if a host wants HCB, it sends an HCB request (Solicit) > > and receives > > HCB and/or ICB replies > > - if a host wants ICB, it sends an ICB request > > (Information-request) > > and receives ICB replies > > > > 1 is a requirement in scenarios with limited resources (e.g., > > wireless), where polling for DHCP is unacceptable. 2 is a > > requirement to avoid timeout delays or other complexity in getting > > ICB reply when > > host would > > prefer HCB reply. > > > > If I've got that right, we can meet the two requirements with a couple > > of small updates to existing specs: > > > > 1) If an RA is received with the M and/or the O bit is set, DHCP > > service is available over the link through which the RA > > was received > > (no differentiation between HCB and ICB DHCP) > > > > 2) If a DHCP server receives an HCB request (Solicit) but can only > > supply an ICB, the server can respond with the ICB reply (note that > > according RFC 3115, the server would respond with an "HCB-nak" > > [Advertise containing only an error code]) > > > > In addition to meeting the requirements, these updates are mostly (if > > not entirely) operationally compatible with existing clients and > > servers. > > I completely agree with this analysis and support proceeding on this > basis. But it does beg the question, why do we need two bits to signal a > binary condition? > > -- Mat -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
